A marketer or data analyst who collects competitor prices essentially does three things: look up prices and offers, place them side by side according to a fixed criterion, and make the outcome usable for a decision about the company's own pricing. This is a task that largely consists of retrieving and ordering data, and that is precisely the kind of work AI can already take over a large part of today, provided a number of conditions are in place.
Three axes are decisive here: structuredness, volume, and room for judgment. Prices are numbers, offers appear on web pages with a recognizable structure, and the comparison itself follows a criterion that has been established beforehand: the same product category, the same unit, the same period. That makes the task structured enough to automate. The volume also helps: ten competitors, a hundred products, repeated weekly, is exactly the kind of repetitive collection work where a system experiences no fatigue and a human does. And the room for judgment is deliberately limited: an agent applies a fixed comparison logic, it does not devise a new strategy.
The remaining axes confirm this picture. There is no customer contact, no physical action, and the compliance sensitivity is low: it concerns public information that anyone can consult. The cost of errors is moderate — a wrongly compared price leads to a distorted picture, not to a breached contract or an angry customer. Creativity plays a role in interpreting unusual offers (a bundle discount, a temporary promotion, a different unit size), but that is more an exception than a rule.
This is a task for an agent: a system that independently consults price pages, extracts data from text and structure, places it within the established comparison criteria, and delivers the result as an overview. This works under two conditions. First: access to public price data, which means the information must be findable and readable — no login wall, no price given only by phone. Second: comparison criteria that are fixed in advance. Without that agreement — what counts as 'comparable', which unit, which period — an agent produces a collection of numbers without anyone knowing whether they may be placed side by side.
This is also exactly where the picture becomes nuanced. Collecting and ordering is for AI. The translation of that comparison into a pricing decision — raising, lowering, launching a promotion — remains human work, involving judgment about margin, positioning, and customer relations. The task in this scan concerns drawing up the comparison, not the decision that follows it.
An online shop that tracks the prices of fifteen competitors across a hundred product categories on a weekly basis can largely leave that process to an agent: it collects the data, recognizes offers, and puts everything into a fixed overview. A marketer then assesses the outcome — is the comparison correct, is something odd going on with a competitor who is suddenly a third cheaper — and approves it or sends it back with a reason. That is the second category from the taxonomy: AI does the work, a human provides oversight with judgment attached.
At a B2B service provider where prices are only available on request, quotes are custom-made, and 'comparable' has no unambiguous meaning, the picture shifts. Structuredness decreases, room for judgment increases, and the work relies more heavily on an analyst who gathers information by phone or through a network and personally assesses what counts as comparable. That is then less a task for an agent and more a task in which AI supports — for instance by structuring notes — while judgment remains with the human.
This shift affects more places within a company than just price monitoring. Anyone who wants to have advertising budget adjusted based on results sees a comparable pattern: structured data, high volume, limited room for judgment, and an agent that can handle most of the work as long as the criteria are fixed. Anyone looking at how website statistics are analyzed also sees the same combination of collecting, ordering, and only then interpreting. Where more creative work is involved, such as with writing a social media post, the balance between automation and oversight is different again. The pattern is always the same: the more structured the input and the higher the volume, the sooner an agent can do the bulk of the work, with a human assessing the outcome rather than doing the work itself.
This is not a statement about the position of marketer or data analyst, and not a basis for building personnel decisions on. Decisions about positions and staffing levels are subject to their own legal requirements, separate from what a task analysis like this one shows. This page describes a task, not a role, and a role usually consists of much more than one task.
To see how this applies to your own company, the free quickscan is a first step: twelve questions, no account required, resulting in an indication of what portion of the hours in this kind of work can be taken over by AI today. The full work scan, which breaks down your company's work into tasks along these eight axes, is still under construction and will be offered here later.
Vraag maar. Ik ken de kennisbank van deze site; wat ik niet weet, zeg ik erbij.
Answers come from this site’s knowledge base. Not tailored advice, and not a scan of your company.